Beta values for hyper- and hypo- methylated probes

Question

[rasii]

Hi all,

I am working on some DNA methylation data and I have used RnBeads to analyse it. I just wanted to know what people have used to classify differentially methylated probes/regions that are hypermethylated, hypomethylated and unmethylated?

From the beta values which is between 0 - 1, what range falls under the above mentioned categories? Can i use >0.8 as hypermethy, <0.2 as hypometh, and what would be methylated and unmethylated.

Is there a paper that gives details of this?

Thanks

[archi]

Hi Rasii,

Could you please guide me how to get normalised beta value matrix? I got 9*3 = 27 samples. ( 9 condition , 3 replicate each). Thanks!

[archi]

Hi Nick and Rasii,

I have got WGBS done on 5 condition with 3 replicates each. I want to compare pattern of methylation among condition for all CpG sites. For this I want to make a matrix of beta value for each CpG site for each condition. And then do WGCNA to cluster them in different module.

Could you please suggest me how can I make this matrix and do normalisation please? Can I also do the matrix for probes, probe would be 100bp window of genome?

Or is there any other way I can do this comparison?

Many thanks

Answer 1

[Chip]

I think you are mixing different concepts.

Methylated and unmethylated generally refer to single sites. When you examine a bulk of cells, the beta values represent the fraction of alleles that are methylated. If you get beta=1 for a given site you know that site is always methylated in that gruop of cells (and always unmethylated if beta=0).

Hypermethylation and hypomethylation are used when you compare different groups (time points, tumor vs normal ecc.). You can have an hypermethylation even with a mean beta=0.5 if the control group has a mean beta=0.

[rasii]

Thanks Nick it makes sense now :)